Abstract Water and gas production and potential water treatment facility requirements for the Marcellus formation are discussed using data analytic methods. These methods aim to handle dataset diversity and scale, and apply data analytics for statistical imputation, estimating future drilling activity, fluids production, and the optimization of water recycling facility locations and size. The objective of this study is to quantify and predict the volumes of produced fluids in the short- and medium-term for the Marcellus shale. The paper accomplishes this objective for the Pennsylvania section comprising 10,000 wells. The application of data analytics to large-scale, data-intensive, low-integrity public environmental databases is illustrated, and challenges of implementation methods are discussed and resolved. In addition, a special class of data analytic tools and workflows for spatiotemporal analysis (spatially correlated variation of parameters with time) is discussed and implemented. The results quantify the prospect of future drilling activity, and water and gas production for all Pennsylvania counties in the Marcellus. Finally, several practical problems of interest on applications of predictive analytics and management support are proposed and solved. The limitations of the proposed workflow are briefly discussed.

中文翻译：

---

预测 Marcellus 页岩采出水的数据分析工作流程

摘要 使用数据分析方法讨论了 Marcellus 地层的水和气生产以及潜在的水处理设施要求。这些方法旨在处理数据集的多样性和规模，并将数据分析应用于统计插补、估计未来的钻井活动、流体生产以及水回收设施位置和规模的优化。本研究的目的是量化和预测 Marcellus 页岩的短期和中期产出流体量。该论文为包含 10,000 口井的宾夕法尼亚部分实现了这一目标。阐述了数据分析在大规模、数据密集型、低完整性公共环境数据库中的应用，并讨论和解决了实施方法的挑战。此外，讨论并实施了一类特殊的数据分析工具和工作流程，用于时空分析（参数随时间的空间相关变化）。结果量化了未来钻井活动的前景，以及马塞勒斯所有宾夕法尼亚县的水和天然气生产。最后，提出并解决了有关预测分析和管理支持应用的几个实际问题。简要讨论了建议工作流程的局限性。提出并解决了关于预测分析和管理支持应用的几个实际问题。简要讨论了建议工作流程的局限性。提出并解决了关于预测分析和管理支持应用的几个实际问题。简要讨论了建议工作流程的局限性。